|
1 # -*- coding: utf-8 -*- |
|
2 |
|
3 # Copyright (c) 2012 Detlev Offenbach <detlev@die-offenbachs.de> |
|
4 # |
|
5 |
|
6 """ |
|
7 Module implementing some FTP related utilities. |
|
8 """ |
|
9 |
|
10 from PyQt4.QtCore import QObject, QDate, QDateTime, QTime |
|
11 from PyQt4.QtNetwork import QUrlInfo |
|
12 |
|
13 |
|
14 class FtpDirLineParserError(Exception): |
|
15 """ |
|
16 Exception class raised, if a parser issue was detected. |
|
17 """ |
|
18 |
|
19 |
|
20 class FtpDirLineParser(QObject): |
|
21 """ |
|
22 Class to parse lines returned by a FTP LIST command. |
|
23 """ |
|
24 MonthnamesNumbers = { |
|
25 "jan": 1, |
|
26 "feb": 2, |
|
27 "mar": 3, |
|
28 "apr": 4, |
|
29 "may": 5, |
|
30 "jun": 6, |
|
31 "jul": 7, |
|
32 "aug": 8, |
|
33 "sep": 9, |
|
34 "oct": 10, |
|
35 "nov": 11, |
|
36 "dec": 12, |
|
37 } |
|
38 |
|
39 UnixMode = 0 |
|
40 WindowsMode = 1 |
|
41 MacMode = 2 |
|
42 |
|
43 def __init__(self, parent=None): |
|
44 """ |
|
45 Constructor |
|
46 |
|
47 @param parent reference to the parent object (QObject) |
|
48 """ |
|
49 super().__init__(parent) |
|
50 |
|
51 self.__parseLine = self.__parseUnixLine |
|
52 self.__modeSwitchAllowed = True |
|
53 |
|
54 def __ignoreLine(self, line): |
|
55 """ |
|
56 Private method to check, if the line should be ignored. |
|
57 |
|
58 @param line to check (string) |
|
59 @return flag indicating to ignore the line (boolean) |
|
60 """ |
|
61 return line.strip() == "" or \ |
|
62 line.strip().lower().startswith("total ") |
|
63 |
|
64 def __parseUnixMode(self, modeString, urlInfo): |
|
65 """ |
|
66 Private method to parse a Unix mode string modifying the |
|
67 given URL info object. |
|
68 |
|
69 @param modeString mode string to be parsed (string) |
|
70 @param urlInfo reference to the URL info object (QUrlInfo) |
|
71 @exception FtpDirLineParserError Raised if the mode cannot be parsed. |
|
72 """ |
|
73 if len(modeString) != 10: |
|
74 raise FtpDirLineParserError("invalid mode string '{0}'".format(modeString)) |
|
75 |
|
76 modeString = modeString.lower() |
|
77 |
|
78 permission = 0 |
|
79 if modeString[1] != '-': |
|
80 permission |= QUrlInfo.ReadOwner |
|
81 if modeString[2] != '-': |
|
82 permission |= QUrlInfo.WriteOwner |
|
83 if modeString[3] != '-': |
|
84 permission |= QUrlInfo.ExeOwner |
|
85 if modeString[4] != '-': |
|
86 permission |= QUrlInfo.ReadGroup |
|
87 if modeString[5] != '-': |
|
88 permission |= QUrlInfo.WriteGroup |
|
89 if modeString[6] != '-': |
|
90 permission |= QUrlInfo.ExeGroup |
|
91 if modeString[7] != '-': |
|
92 permission |= QUrlInfo.ReadOther |
|
93 if modeString[8] != '-': |
|
94 permission |= QUrlInfo.WriteOther |
|
95 if modeString[9] != '-': |
|
96 permission |= QUrlInfo.ExeOther |
|
97 urlInfo.setPermissions(permission) |
|
98 |
|
99 if modeString[0] == "d": |
|
100 urlInfo.setDir(True) |
|
101 urlInfo.setFile(False) |
|
102 elif modeString[0] == "l": |
|
103 urlInfo.setSymLink(True) |
|
104 elif modeString[0] == "-": |
|
105 urlInfo.setDir(False) |
|
106 urlInfo.setFile(True) |
|
107 |
|
108 def __parseUnixTime(self, monthAbbreviation, day, yearOrTime, urlInfo): |
|
109 """ |
|
110 Private method to parse a Unix date and time indication modifying |
|
111 the given URL info object. |
|
112 |
|
113 |
|
114 Date time strings in Unix-style directory listings typically |
|
115 have one of these formats: |
|
116 <ul> |
|
117 <li>"Nov 23 02:33" (month name, day of month, time)</li> |
|
118 <li>"May 26 2005" (month name, day of month, year)</li> |
|
119 </ul> |
|
120 |
|
121 @param monthAbbreviation abbreviation of the month name (string) |
|
122 @param day day of the month (string) |
|
123 @param yearOrTime string giving the year or a time (string) |
|
124 @param urlInfo reference to the URL info object (QUrlInfo) |
|
125 @exception FtpDirLineParserError Raised if the month abbreviation is |
|
126 not recognized. |
|
127 """ |
|
128 try: |
|
129 month = FtpDirLineParser.MonthnamesNumbers[monthAbbreviation.lower()] |
|
130 except KeyError: |
|
131 raise FtpDirLineParserError("illegal month abbreviation '{0}'".format( |
|
132 monthAbbreviation)) |
|
133 day = int(day) |
|
134 if ':' in yearOrTime: |
|
135 year = QDate.currentDate().year() |
|
136 hour, minute = yearOrTime.split(':') |
|
137 hour = int(hour) |
|
138 minute = int(minute) |
|
139 else: |
|
140 year = int(yearOrTime) |
|
141 hour = 0 |
|
142 minute = 0 |
|
143 |
|
144 lastModified = QDateTime(QDate(year, month, day), QTime(hour, minute)) |
|
145 urlInfo.setLastModified(lastModified) |
|
146 |
|
147 def __splitUnixLine(self, line): |
|
148 """ |
|
149 Split a line of a Unix like directory listing into meta data, |
|
150 number of links, user, group, size, month, day, year or time |
|
151 and name. |
|
152 |
|
153 @return tuple of nine strings giving the meta data, |
|
154 number of links, user, group, size, month, day, year or time |
|
155 and name |
|
156 @exception FtpDirLineParserError Raised if the line is not of a |
|
157 recognized Unix format. |
|
158 """ |
|
159 # This method encapsulates the recognition of an unusual |
|
160 # Unix format variant. |
|
161 lineParts = line.split() |
|
162 fieldCountWithoutUserID = 8 |
|
163 fieldCountWithUserID = fieldCountWithoutUserID + 1 |
|
164 if len(lineParts) < fieldCountWithoutUserID: |
|
165 raise FtpDirLineParserError("line '{0}' cannot be parsed".format(line)) |
|
166 |
|
167 # If we have a valid format (either with or without user id field), |
|
168 # the field with index 5 is either the month abbreviation or a day. |
|
169 try: |
|
170 int(lineParts[5]) |
|
171 except ValueError: |
|
172 # Month abbreviation, "invalid literal for int" |
|
173 lineParts = line.split(None, fieldCountWithUserID - 1) |
|
174 else: |
|
175 # Day |
|
176 lineParts = line.split(None, fieldCountWithoutUserID - 1) |
|
177 userFieldIndex = 2 |
|
178 lineParts.insert(userFieldIndex, "") |
|
179 |
|
180 return lineParts |
|
181 |
|
182 def __parseUnixLine(self, line): |
|
183 """ |
|
184 Private method to parse a Unix style directory listing line. |
|
185 |
|
186 @param line directory line to be parsed (string) |
|
187 @return URL info object containing the valid data (QUrlInfo) |
|
188 @exception FtpDirLineParserError Raised if the line is not of a |
|
189 recognized Unix format. |
|
190 """ |
|
191 modeString, nlink, user, group, size, month, day, \ |
|
192 yearOrTime, name = self.__splitUnixLine(line) |
|
193 |
|
194 urlInfo = QUrlInfo() |
|
195 self.__parseUnixMode(modeString, urlInfo) |
|
196 self.__parseUnixTime(month, day, yearOrTime, urlInfo) |
|
197 urlInfo.setOwner(user) |
|
198 urlInfo.setGroup(group) |
|
199 urlInfo.setSize(int(size)) |
|
200 name = name.strip() |
|
201 i = name.find(" -> ") |
|
202 if i >= 0: |
|
203 name = name[:i] |
|
204 urlInfo.setName(name) |
|
205 |
|
206 return urlInfo |
|
207 |
|
208 def __parseWindowsTime(self, date, time, urlInfo): |
|
209 """ |
|
210 Private method to parse a Windows date and time indication modifying |
|
211 the given URL info object. |
|
212 |
|
213 Date time strings in Windows-style directory listings typically |
|
214 have the format "10-23-12 03:25PM" (month-day_of_month-two_digit_year, |
|
215 hour:minute, am/pm). |
|
216 |
|
217 @param date date string (string) |
|
218 @param time time string (string) |
|
219 @param urlInfo reference to the URL info object (QUrlInfo) |
|
220 @exception FtpDirLineParserError Raised if either of the strings is not |
|
221 recognized. |
|
222 """ |
|
223 try: |
|
224 month, day, year = [int(part) for part in date.split('-')] |
|
225 if year >= 70: |
|
226 year = 1900 + year |
|
227 else: |
|
228 year = 2000 + year |
|
229 except (ValueError, IndexError): |
|
230 raise FtpDirLineParserError("illegal date string '{0}'".format(month)) |
|
231 try: |
|
232 hour, minute, am_pm = time[0:2], time[3:5], time[5] |
|
233 hour = int(hour) |
|
234 minute = int(minute) |
|
235 except (ValueError, IndexError): |
|
236 raise FtpDirLineParserError("illegal time string '{0}'".format(month)) |
|
237 if hour == 12 and am_pm == 'A': |
|
238 hour = 0 |
|
239 if hour != 12 and am_pm == 'P': |
|
240 hour = hour + 12 |
|
241 |
|
242 lastModified = QDateTime(QDate(year, month, day), QTime(hour, minute)) |
|
243 urlInfo.setLastModified(lastModified) |
|
244 |
|
245 def __parseWindowsLine(self, line): |
|
246 """ |
|
247 Private method to parse a Windows style directory listing line. |
|
248 |
|
249 @param line directory line to be parsed (string) |
|
250 @return URL info object containing the valid data (QUrlInfo) |
|
251 @exception FtpDirLineParserError Raised if the line is not of a |
|
252 recognized Windows format. |
|
253 """ |
|
254 try: |
|
255 date, time, dirOrSize, name = line.split(None, 3) |
|
256 except ValueError: |
|
257 # "unpack list of wrong size" |
|
258 raise FtpDirLineParserError("line '{0}' cannot be parsed".format(line)) |
|
259 |
|
260 urlInfo = QUrlInfo() |
|
261 self.__parseWindowsTime(date, time, urlInfo) |
|
262 if dirOrSize.lower() == "<dir>": |
|
263 urlInfo.setDir(True) |
|
264 urlInfo.setFile(False) |
|
265 else: |
|
266 urlInfo.setDir(False) |
|
267 urlInfo.setFile(True) |
|
268 try: |
|
269 urlInfo.setSize(int(dirOrSize)) |
|
270 except ValueError: |
|
271 raise FtpDirLineParserError("illegal size '{0}'".format(dirOrSize)) |
|
272 urlInfo.setName(name) |
|
273 |
|
274 return urlInfo |
|
275 |
|
276 def parseLine(self, line): |
|
277 """ |
|
278 Private method to parse a directory listing line. |
|
279 |
|
280 This implementation support Unix and Windows style directory |
|
281 listings. It tries Unix style first and if that fails switches |
|
282 to Windows style. If that fails as well, an exception is raised. |
|
283 |
|
284 @param line directory line to be parsed (string) |
|
285 @return URL info object containing the valid data (QUrlInfo) |
|
286 @exception FtpDirLineParserError Raised if the line is not of a |
|
287 recognized format. |
|
288 """ |
|
289 if self.__ignoreLine(line): |
|
290 return None |
|
291 |
|
292 try: |
|
293 return self.__parseLine(line) |
|
294 except FtpDirLineParserError: |
|
295 if self.__modeSwitchAllowed: |
|
296 self.__parseLine = self.__parseWindowsLine |
|
297 self.__modeSwitchAllowed = False |
|
298 return self.__parseLine(line) |
|
299 else: |
|
300 raise |